Understanding LLM Embeddings for Regression

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tang, Eric, Yang, Bangding, Song, Xingyou
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912232777973760
author Tang, Eric
Yang, Bangding
Song, Xingyou
author_facet Tang, Eric
Yang, Bangding
Song, Xingyou
contents With the rise of large language models (LLMs) for flexibly processing information as strings, a natural application is regression, specifically by preprocessing string representations into LLM embeddings as downstream features for metric prediction. In this paper, we provide one of the first comprehensive investigations into embedding-based regression and demonstrate that LLM embeddings as features can be better for high-dimensional regression tasks than using traditional feature engineering. This regression performance can be explained in part due to LLM embeddings over numeric data inherently preserving Lipschitz continuity over the feature space. Furthermore, we quantify the contribution of different model effects, most notably model size and language understanding, which we find surprisingly do not always improve regression performance.
format Preprint
id arxiv_https___arxiv_org_abs_2411_14708
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Understanding LLM Embeddings for Regression
Tang, Eric
Yang, Bangding
Song, Xingyou
Machine Learning
Artificial Intelligence
Computation and Language
With the rise of large language models (LLMs) for flexibly processing information as strings, a natural application is regression, specifically by preprocessing string representations into LLM embeddings as downstream features for metric prediction. In this paper, we provide one of the first comprehensive investigations into embedding-based regression and demonstrate that LLM embeddings as features can be better for high-dimensional regression tasks than using traditional feature engineering. This regression performance can be explained in part due to LLM embeddings over numeric data inherently preserving Lipschitz continuity over the feature space. Furthermore, we quantify the contribution of different model effects, most notably model size and language understanding, which we find surprisingly do not always improve regression performance.
title Understanding LLM Embeddings for Regression
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2411.14708